Ensemble Fusion of Classifiers with Kernel PCA for Breast Cancer Classification

Abstract

Breast cancer, predominantly affecting females, ranks as the most prevalent cancer among women globally, with potentially fatal consequences. Its invasive nature poses a significant health threat. Delayed diagnosis due to asymptomatic early stages hinders effective medical intervention. Early screenings prove pivotal in reducing breast cancer mortality. Beyond conventional diagnostic approaches, machine learning employs health data to predict breast cancer risk. This study employs Wisconsin breast cancer diagnosis data from the UCI machine learning repository. Class Imbalance is handled using AllKNN under-sampling technique and feature extraction using Kernel Principal Component Analysis (KPCA) were conducted. Logistic Regression, Support Vector Machine and Ensemble Learning (majority voting) are proposed for achieving high predictive accuracy.